Characteristic activities of slow earthquakes in Japan

Abstract

Slow earthquakes are a recently discovered phenomenon that mainly occur updip and downdip of the seismogenic zones of great earthquakes along the subducting plate interface. The spatiotemporal activity of various slow earthquakes occurring in the Nankai subduction zone is characterized by along-strike heterogeneity and along-dip systematic changes. Various slow earthquakes are horizontally distributed at their own depths and along-strike segments can be observed with respect to this distribution downdip of the locked zone; however, slow and great earthquakes occur in the same depth range near the Nankai Trough and Japan Trench axes. The frequently observed spatiotemporal interactions between different slow earthquakes can be attributed to their sensitivity and the stress transfer of the surrounding areas. This stress transfer is expected to extend to the adjacent sections in the seismogenic zone. Therefore, precise monitoring of slow earthquakes is important for future evaluations of great earthquakes, which requires the long-term maintenance and continuous improvement of the high-quality observation networks.

1. Introduction

The term “slow earthquake” is used for various types of fault slip phenomena having longer characteristic times than those of regular earthquakes. Two types of slow earthquakes, i.e., long-term slow slip events (SSEs) and low-frequency tremor (see Table 1), were initially discovered in southwest Japan, just downdip of the seismogenic zone of great earthquakes along the subducting Philippine Sea plate boundary, by Hirose et al.¹⁾ and Obara,²⁾ respectively. Subsequently, various types of slow earthquakes have been detected in many tectonic settings around the world, especially the subduction zones along the Pacific Rim (e.g., Cascadia, Mexico, Alaska, New Zealand, Costa Rica, and Chile). Slow earthquakes have also been observed along the strike-slip transform fault systems, such as the San Andreas Fault,³⁾ and along an arc-continent collisional zone, such as beneath the Central Range in Taiwan.⁴⁾

Table 1. Glossary of the terms for slow earthquakes used in this paper

Low-frequency tremor

Weak seismic vibrations lasting from minutes to days with a predominant frequency of 1–10 Hz, which is lower than that of a regular microearthquake having a comparable amplitude to that of tremor. Tremor comprises swarm activities of LFE.

Low-frequency earthquake (LFE)

A seismic event radiating a waveform with a predominant frequency of 1–10 Hz, which is lower than that of a regular earthquake having a comparable amplitude. LFEs are mainly distributed in volcanic areas and subducting plate interfaces. The LFEs along the subducting plate interface are considered an elementary process of low-frequency tremor.

Very-low-frequency earthquake (VLFE)

A seismic event radiating a waveform with a predominant period of 10–100 s associated with very weak amplitudes of high-frequency components.

Slow slip event (SSE)

A transient slip along the fault plane with no radiation of a seismic wave.

Long-term SSE

SSEs with a duration ranging from months to years.

Short-term SSE

SSEs with a duration ranging from days to weeks.

Episodic tremor and slip (ETS)

Simultaneous occurrence of short-term SSEs, VLFEs, and tremor.

Friction experiments⁵⁾ and thermal modeling studies⁶⁾ suggest that the source region of the slow earthquake corresponds to the transition zone between the locked and stable sliding zones of the plate boundary fault. This transition zone is also expected updip of the locked zone based on the friction modeling of the subducting plate interface.⁵⁾ Shallow slow earthquakes have been detected in this shallow transition zone using inland broadband seismograph networks⁷⁾ and ocean-bottom observation instruments.⁸⁾ Therefore, slow earthquakes are a phenomenon related to the transition from fast coseismic slip to stable sliding along the subducting plate interface.

This paper presents an overview of the slow earthquake phenomenon in Japan and summarizes the development and deployment of observational networks as the background for the discovery of slow earthquakes. Slow earthquakes are classified into various types based on their characteristic timescale and depth relative to those associated with the seismogenic zone of great earthquakes. Further, each type of slow earthquake is explained based on its basic characteristic activity and discovery process. The spatiotemporal activity patterns used to characterize the various types of slow earthquakes are also discussed with respect to the along-strike segmentation and along-dip systematic change. Finally, the spatiotemporal interactions between different types of slow earthquakes may be related to the interactions between slow and great earthquakes; hence, the spatiotemporal interactions are discussed in this study.

2. Discovery of slow earthquakes in Japan

The discovery of slow earthquakes can be related to the development and deployment of improved observation networks. The development of two observation networks in Japan, i.e., the high-sensitivity seismograph network (Hi-net) operated by the National Research Institute for Earth Science and Disaster Resilience (NIED)⁹⁾ and the global navigation satellite system (GNSS) earth observation network system (GEONET) operated by the Geospatial Information Authority of Japan (GSI),¹⁰⁾ considerably contributed to the scientific study of slow earthquakes. Both the aforementioned observation networks were installed across the Japanese islands after the Kobe earthquake of 1995, and the average spacing between two stations was 20–30 km. Such dense observation networks considerably improve the signal coherency and detection capability of slow earthquakes. The data are transmitted in real time and stored on data servers that evolve with the progress of the data transmission and storage technology. Each Hi-net seismic station includes a set of high-sensitivity seismometers and accelerometers encased in a sensor capsule and installed at the base of a borehole. The accelerometer serves as a broadband seismometer and tiltmeter.¹¹⁾ The coupling of such a broad observation window with a high signal-to-noise ratio (because of the low levels of cultural noise in deep boreholes) is useful to detect various types of slow earthquakes having different characteristic timescales and weak amplitudes.

Two advances in the development of observational instruments have considerably contributed to the recent discovery of shallow slow earthquakes in an offshore environment. One is the improvement of temporarily deployed ocean-bottom observation instruments equipped with broadband sensors and long-life batteries for increasing the spectrum window and observation time window, respectively. The second advancement is the construction of a cabled seafloor observational system, in which the stations are connected using an optical fiber transmitting data in real time to an inland data center. The Dense Ocean floor Network system for Earthquakes and Tsunamis (DONET), comprising approximately 50 observation points, was constructed by the Japan Agency for Marine-Earth Science and Technology (JAMSTEC) offshore to the southeast of the Kii Peninsula for evaluating the potential future great earthquakes in Nankai. The seafloor observation network for earthquakes and tsunamis along the Japan Trench (S-net), which comprises approximately 200 seismometers over the entire offshore region to the east of northern Japan (1000 km N–S and 300 km E–W), was constructed by NIED after the Tohoku earthquake in 2011. The development and deployment of such a seafloor observation system contribute to the monitoring of slow earthquakes and forecasting of the strong ground motion and tsunami generated because of great earthquakes.

3. Classification of slow earthquakes

Slow earthquakes include various fault slip phenomena having different characteristic timescales. The combination of different types of slow earthquakes and their associated activity patterns vary along many subduction zones along the Pacific Rim (Fig. 1).¹²⁾ Therefore, the different characteristics of slow earthquakes may be used to characterize each subduction system. The Nankai subduction zone in Japan is the region at which most types of slow earthquakes can be observed (Fig. 2). Therefore, the observations obtained from the Nankai subduction zone are used to classify slow earthquakes in this study. Almost all the slow earthquakes are detected at the updip and downdip extensions of the locked seismogenic zone on the subducting plate interface. These slow earthquakes occur over a broad range of characteristic timescales ranging from seconds to years. Therefore, the depth of a given slow earthquake relative to that of the locked zone and the characteristic timescale are used to classify slow earthquakes (Fig. 3). The slow earthquakes are classified as deep or shallow depending on whether they are detected below or above the locked zone. In addition, they are classified as seismic or geodetic based on whether they are seismically detected as ground shaking using seismometers or geodetically detected as crustal deformation using GNSSs, strainmeters, or tiltmeters, respectively. The seismic slow earthquakes are mainly classified as low-frequency tremor, with a predominant frequency of several hertz, and very-low-frequency earthquakes (VLFEs), with a predominant period of several tens of seconds. The geodetic slow earthquakes are mainly divided into long-term SSEs, with a typical duration of several months to years, and short-term SSEs, with a typical duration of some days to weeks (Fig. 4).

Fig. 1.

Slow and fast earthquake activity along the Pacific Rim. The characteristic slow earthquake activity (tremor (LFEs), VLFEs, and SSEs) and regular earthquake activity (megathrust earthquakes (MegaEQ) and swarm) along each major fault system are indicated. Everything! indicates that three types of slow earthquakes (tremor, VLFEs, and SSEs) are detected. Blue lines indicate the tectonic plate boundaries. The background image shows the seafloor age.¹²⁾

Fig. 2.

Map of various slow earthquake occurrences in southwest Japan. Slow earthquakes surround the seismogenic zone of great earthquakes in the Nankai Trough region (red polygons). The red dots indicate deep low-frequency tremor, which are usually accompanied by short-term slow slip events (SSEs) and deep very-low-frequency earthquakes (VLFEs). The simultaneous occurrence of the aforementioned three types of slow earthquakes is called episodic tremor and slip (ETS). The yellow ellipses denote the long-term SSE source regions between the ETS and seismogenic zones. The purple ellipses show shallow VLFEs between the seismogenic zone and Nankai Trough axis. The Boso SSEs occur toward the east of the Boso Peninsula near the Sagami Trough and are associated with a regular earthquake swarm activity indicated by the light-blue ellipse.

Fig. 3.

Classification of various slow earthquakes along the subducting plate interface in the Nankai subduction zone and observation instruments deployed in the inland and offshore areas. Slow earthquakes are spatially separated into updip and downdip sections relative to the seismogenic locked zone of great earthquakes and are mainly classified into geodetic and seismic phenomena in the spectral domain. The geodetic slow earthquakes are separated into long- and short-term slow slip events (SSEs), with durations ranging from months to years and days to weeks, respectively. The seismic slow earthquakes are separated into very-low-frequency earthquakes (VLFEs) and low-frequency tremor, with dominant frequencies of 0.01–0.1 Hz and 1–10 Hz, respectively. SSEs have recently been detected in the shallow section of the subduction zone.

Fig. 4.

Example of the records observed from various slow earthquakes. (a) GNSS displacement time series of the long-term SSE detected in the Bungo Channel by GSI GEONET. (b) Ground tilt record of the short-term SSE detected in western Shikoku observed using a NIED borehole tiltmeter. (c) Displacement seismogram of the deep VLFE recorded by the NIED Hi-net high-sensitivity accelerometer with a passband frequency of 0.01 to 0.1 Hz. (d) Velocity seismogram of the tremor recorded by the Hi-net seismometer. (e) Velocity seismogram of the M 3.8 regular earthquake recorded by the Hi-net seismometer.

3-1. Deep slow earthquakes.

In the Nankai Trough region, long-term SSEs are distributed next to the locked zone at the downdip extension of the seismogenic zone. The low-frequency tremor, VLFEs, and short-term SSEs are detected at the deepest section of the deep transition zone. These three phenomena usually occur simultaneously and are termed as “episodic tremor and slip (ETS)”.¹³⁾ The Boso SSEs, which are located at the plate boundary at which the Philippine Sea plate subducts from the Sagami Trough, can be observed at slightly shallower depths when compared with other deep slow earthquakes in the Nankai subduction zone; however, they are defined as deep slow earthquakes because they are located downdip of the Kanto earthquake seismogenic zone.

3-1-1. Low-frequency tremor.

Deep low-frequency tremor was initially discovered in 2002 in southwest Japan²⁾ because of the improved capability of the newly installed Hi-net in detecting weak shaking phenomena. The tremor exhibited a predominant frequency of 1–10 Hz, lasted for minutes to days, and demonstrated no clear P- or S-wave onset. The waveform pattern was similar to that of volcanic tremor; however, this low-frequency tremor was detected in nonvolcanic areas. Therefore, this tremor was originally termed “nonvolcanic tremor”.²⁾ This tremor has since been termed as “low-frequency tremor” (because it comprises many low-frequency earthquakes (LFEs)) or “tectonic tremor” (because it reflects a large-scale tectonic activity related to plate subduction).

Tremor is a superposition of the swarm activity obtained from many LFEs.¹⁴⁾ The envelope correlation method was developed to locate the tremor source based on the coherent envelope pattern of the tremor at neighboring stations because it is difficult to precisely identify the onset of the tremor owing to the unclear low-frequency onset of the LFEs.²⁾ Several researchers have improved the envelope-based methods used to estimate the radiation energy¹⁵⁾ and duration.¹⁶⁾ The Japan Meteorological Agency (JMA) has been applying an ordinary hypocenter determination method by selecting onset phases to develop the LFE catalog.¹⁷⁾ A matched filter method in which the waveforms of the previously detected LFEs are considered to be template events is frequently applied to detect tremor.¹⁴⁾ This method has also been used by JMA to detect LFEs since 2018.

The tremor epicenters are distributed along the depth contours of the subducting Philippine Sea plate, spanning a zone with an approximate area of 600 km × <50 km from the southern Nagano Prefecture to the Bungo Channel. This belt-like tremor zone is separated into several segments. Generally, a swarm-like tremor activity occurs episodically at an approximately regular recurrence interval in case of each segment.¹⁸⁾ Such active tremor episodes are associated with short-term SSEs,¹⁹⁾ which are described in the section 3-1-3. The focal mechanism of the tremor is unclear because of the absence of clear onset. However, based on stacking analysis for LFEs, the focal mechanism of the LFEs is consistent with reverse faulting along the subducting plate interface.²⁰⁾ The tremor episodes usually migrate within each segment at approximately 10 km/day parallel to the strike of the plate geometry,²⁾ and the migration rarely extends to the neighboring segments. Other tremor migration modes can also be observed because the active tremor episode associated with along-strike migration at approximately 10 km/day is occasionally accompanied by the along-strike migration of the tremor propagating backward at order of 100 km/day (rapid tremor reversal (RTR))²¹⁾ and the along-dip migration at order of 1000 km/day.^22),23)

One of the important characteristics of tremor is that it can be triggered by external stress changes. This suggests that tremor is very sensitive. There are two distinct triggering phenomena. One such phenomenon is triggering by seismic wave originating from other earthquakes. The surface waves propagating from large teleseismic earthquakes frequently trigger tremor.²⁴⁾ The tremor bursts triggered by the surface waves are repeatedly activated at an interval of several tens of seconds, which is associated with each phase of the long-period surface waves. The triggered tremor is easier to detect than the ambient tremor because we can focus only on short time windows that include the surface wave packets from teleseismic earthquakes; further, we attempt to estimate the enhancement of the coherent amplitude corresponding to the surface wave phases. Both long-period surface waves and high-frequency body waves trigger tremor. Some tremor bursts are observed approximately 5 minutes after the arrival of the body waves originating from intraslab earthquakes occurring just beneath the triggered tremor activity.²⁵⁾ Tidal modulation is another triggering phenomenon. Repeated tremor activation at 12- or 24-h intervals is commonly observed during active tremor episodes associated with short-term SSEs.²⁶⁾

3-1-2. Deep VLFEs.

Deep VLFEs, with a predominant period of several tens of seconds, were discovered in 2007 based on their association with active ETS²⁷⁾ by applying the GRiD-MT (Grid-based Realtime Determination of Moment Tensors) method.²⁸⁾ Further, the focal mechanism of VLFEs was estimated via centroid moment tensor analysis to obtain a reverse fault solution consistent with the geometry and motion of the subducting Philippine Sea plate. This is one of the key lines of evidence that prove that ETS is an interplate shear-slip phenomenon. VLFEs are mainly events with M_W < 4. Deep VLFEs are heterogeneously concentrated within areas in the belt-like tremor source zone,²⁹⁾ and tremor is homogeneously distributed across this zone.¹⁸⁾ The stacking analysis of broadband seismograms with reference to the tremor time revealed that the VLFE distribution, including weak-amplitude events, covered the entire tremor source zone.³⁰⁾ Deep VLFEs have been comprehensively detected in western Shikoku, and long-term temporal variations have been observed with respect to the VLFE seismicity³¹⁾ via matched filter analysis based on the synthetic waveforms calculated from the focal mechanisms of deep VLFEs.³⁰⁾

3-1-3. Short-term SSEs.

SSEs were discovered in Cascadia in 2001 using the regional GPS network,³²⁾ and an association between SSEs and tremor episodes was discovered in 2003.¹³⁾ The discovery of ETS in Cascadia resulted in the expectation that similar SSEs associated with tremor would occur in southwest Japan, and ETS was discovered at western Shikoku in 2004.¹⁹⁾ Finally, a tiny crustal deformation was detected in southwest Japan using tiltmeters.¹¹⁾ These SSEs had durations ranging from some days to a week and were associated with active tremor episodes. The slip parameters were estimated to denote reverse faulting along the upper boundary of the subducting Philippine Sea plate,¹⁹⁾ which is the same focal mechanism as that of the interplate megathrust earthquakes. The duration was considerably less than those of the previously discovered SSEs in the Bungo Channel,¹⁾ which are mentioned in the next section. Therefore, the SSEs associated with tremor have been termed “short-term SSEs” and the Bungo Channel SSEs have been termed “long-term SSEs”,³³⁾ to distinguish between SSEs with different timescales. The short-term SSEs in southwest Japan were initially detected using the Hi-net borehole tiltmeter because of the smaller magnitude of these SSEs compared with those in Cascadia. Subsequently, they were detected using borehole strainmeters operated by JMA and the National Institute of Advanced Industrial Science and Technology. Short-term SSEs have since been detected by GEONET after the development of a refined detection method and achieving improvements in solving the coordinate value.³⁴⁾

The estimated source fault of short-term SSEs approximately corresponds to the tremor distribution for each episode. Short-term SSEs occur periodically at a nearly constant recurrence interval associated with the tremor episode at each ETS segment. The SSEs in western Shikoku occur at an approximately 6-month interval, with these approximately M_W 6 events resulting in 1–2 cm of slip.¹⁹⁾ The spatial distribution of SSEs usually migrates with tremor. The spatiotemporal evolution analysis of short-term SSEs indicates that the tremor epicenters are well collocated with the peaks in slip distribution.³⁵⁾ This observation is similar to that in Cascadia.³⁶⁾ Therefore, tremor is caused by a seismic failure at small seismic patches embedded on the short-term SSE fault plane because of stress loading by the surrounding transient slip. This model suggests that the migration of the tremor may reflect the propagation of the short-term SSE rupture front³⁶⁾ and supports that tremor is a proxy for SSEs.³⁷⁾ Tremor is the easiest slow earthquake to detect among the ETS phenomena because of its properties and detectability in observational networks; therefore, tremor is frequently treated as a proxy for ETS.

3-1-4. Long-term SSEs.

Long-term SSEs were discovered in the Bungo Channel¹⁾ and Tokai region³⁸⁾ shortly after the deployment of GEONET in Japan. Long-term SSEs have since been detected at other locations, including the Kii Channel,³⁹⁾ central Shikoku,^40),41) northern Miyazaki Prefecture,^42),43) and southern Miyazaki Prefecture.⁴⁴⁾ These long-term SSEs are commonly observed between the locked and ETS zones along the subducting plate interface in southwest Japan.

Each long-term SSE has its own unique property (recurrence interval, duration, and size) in each region. The M_W 6.8-class long-term SSEs in the Bungo Channel have a duration of six months and a recurrence interval of 6–7 years (1997, 2003, and 2010). A similar-sized SSE could be observed in 2019; however, two small SSEs occurred in 2014 and 2016 during this inter-SSE period. Therefore, these small SSEs may release a small amount of accumulated stress, delaying the occurrence of large SSEs.

A M_W 7-class long-term SSE was observed from mid-2000 to mid-2005 near Lake Hamana in the Tokai region by GEONET.⁴⁵⁾ The crustal deformation owing to the long-term SSE was detected using the tiltmeter installed at the Mikkabi station. A similar ground tilting was previously observed at the same tilt station from 1989 to 1991.⁴⁶⁾ Furthermore, a small-sized long-term SSE occurred in the same region from 2013 to 2015.⁴⁷⁾ Therefore, the recurrence behavior of the long-term SSEs in Tokai is irregular.

3-1-5. Boso SSEs.

SSEs occur at an approximately 6-year interval along the eastern coastline and offshore area of the Boso Peninsula, central part of Japan. They usually have a size of M_W 6.5 with a duration of approximately one week. The SSE fault is located at a depth of approximately 10–20 km. One of the most characteristic phenomena of the Boso SSEs is their association with regular earthquake swarms, usually including M 4–5-class earthquakes and repeating earthquakes at the downdip edge of the Boso SSE source fault region.⁴⁸⁾ Therefore, the occurrence of M 4–5-class medium-sized earthquakes during swarm activity can be predicted immediately after the detection of the initiation of the Boso SSEs. The recurrence interval of the Boso SSEs decreased gradually before the Tohoku earthquake in 2011; however, the recurrence interval was reset by the Tohoku earthquake.⁴⁸⁾ Repeated earthquakes were detected during the active aftershock period of the Tohoku earthquake in March 2011 even though there was no meaningful geodetic signal owing to the large coseismic slip and afterslip.⁴⁹⁾ Thus, a small SSE may have occurred during the Tohoku earthquake sequence in 2011. Boso SSEs substantially occurred in November 2011, January 2014, and June 2018, and the recurrence interval was observed to gradually increase after the Tohoku earthquake. Precise spatiotemporal evolution analysis revealed that the migration of the earthquake swarms was associated with slip propagation and that the 2014 SSE was preceded by slow slip acceleration that began one month before the rapid acceleration of the SSE.⁵⁰⁾

3-2. Shallow slow earthquakes.

Various slow earthquakes were detected between the updip edge of the seismogenic zone of great earthquakes and the Nankai Trough axis. Shallow slow earthquakes were also detected along the Japan and Kuril trenches and near the Ryukyu Islands. The detection of these shallow slow earthquakes in several offshore regions over the previous decade can be attributed to the recent development of the seafloor observation technology and the deployment of the seafloor observation instruments.

3-2-1. Shallow VLFEs.

Shallow VLFEs were initially reported in 2003 in the offshore region in southwest Japan using the NIED F-net broadband continuous seismograph data,⁵¹⁾ and several spatiotemporal clusters of shallow VLFEs could be observed in the landward section of the Nankai Trough axis via back-projection analysis using the continuous waveform data obtained from the Hi-net high-sensitivity accelerometers.⁷⁾ The discovery of shallow VLFEs led to the discovery of deep VLFEs.²⁷⁾ Subsequently, shallow VLFEs have been detected at the Nankai Trough, Ryukyu Islands,⁵²⁾ offshore Tokachi,⁵³⁾ and Japan Trench.⁵⁴⁾ The focal mechanism of the shallow VLFEs offshore to the southeast of the Kii Peninsula was estimated to denote a reverse fault solution with a steep dip angle compared with that of the plate interface near the trench axis. This was interpreted as a slow rupture along a spray fault and/or out-of-sequence thrust fault within the accretionary prism.⁵⁵⁾

The aforementioned research results were derived using inland broadband observation networks. Recently, shallow VLFEs have been observed using temporal and permanent seafloor observation networks. The shallow VLFEs offshore to the southeast of the Kii Peninsula detected by closely deployed ocean-bottom broadband seismometers indicated reverse fault mechanisms with nearly horizontal fault planes.⁵⁶⁾ Thus, shallow VLFEs are shear-slip phenomena along the decollement or plate boundary. Many shallow VLFEs and tremor have been detected using the DONET cabled seafloor observation system.⁵⁷⁾ One of the important results obtained using DONET is the discovery of slow earthquakes with a predominant period of 1–10 s.⁵⁸⁾ This period corresponds to a spectrum gap between tremor and VLFEs owing to the large amplitude of the microseisms; however, such new events were only observed after conducting a stacking analysis of many tremor and VLFEs activated after the April 2016 M_W 6-class interplate earthquake. This proves that slow earthquakes involve continuous spectrum components.

A matched filter analysis in which the synthetic template events obtained based on the plate geometry and motion has enabled the comprehensive detection of shallow VLFEs in the Japan Trench region.⁵⁹⁾ They revealed temporal activation and inactivation immediately after the 2011 Tohoku earthquake in the afterslip and coseismic slip regions, respectively. The cumulative number of shallow VLFEs in the afterslip region of the 2003 Tokachi-oki earthquake was similar to the crustal deformation pattern that can be attributed to the afterslip.⁵³⁾ These observations suggest that the shallow VLFE activity reflects an interplate transient slip.

Shallow VLFEs are occasionally triggered by other phenomena such as nearby major earthquakes, like as triggered deep tremor. For example, shallow VLFEs were activated in Hyuga-nada after the 2003 Tokachi-oki earthquake and offshore to the southeast of the Kii Peninsula after the occurrence of a M_W 7.5 intraslab earthquake in the same area.⁵³⁾ However, the shallow VLFE seismicity in the Ryukyu Islands was tidally modulated, denoting maximum activation during low tide and a tidal response exhibiting regional and seasonal dependence.⁶⁰⁾

3-2-2. Shallow tremor.

The predominantly high-frequency composition of tremor indicates that it is strongly attenuated with the increasing distance. Consequently, in situ offshore observations are required to detect shallow tremor even though the long-period seismic signal from shallow VLFEs efficiently propagates to distant inland broadband seismic stations. Shallow tremor near the trench axis was initially discovered in 2009 offshore to the southeast of the Kii Peninsula by temporarily deploying ocean-bottom seismometers,⁶¹⁾ and there was no clear evidence with respect to the collocation of the shallow tremor and VLFEs. An active shallow tremor episode lasting for one month was detected based on the temporary offshore seismic observations in Hyuga-nada.⁸⁾ This shallow tremor episode was spatiotemporally consistent with the shallow VLFE activity detected by the inland broadband observation network. This is the first result indicating the coincident occurrence of shallow VLFEs and tremor. Furthermore, the shallow tremor episode included two migration modes with different speeds based on the very precise hypocentral location of the shallow tremor because of the in situ observation area. One is northward migration at approximately 20 km/day, which is similar to the along-strike migration of the deep tremor at approximately 10 km/day. The other is faster migration that propagates backward at order of 100 km/day, which is similar to the RTR detected during deep tremor episodes.²¹⁾ The common characteristics associated with the coincident occurrence of VLFEs and tremor and the existence of multiscale migration modes observed in the shallow and deep sections suggest that seismic slow earthquakes are associated with SSEs in the shallow section, similar to the relation between the seismic slow earthquake phenomenon and short-term SSEs in the deeper section.

Shallow tremor was detected by the ocean-bottom seismometers temporarily deployed near the Japan Trench region before the Tohoku earthquake in 2011.⁶²⁾ The S-net seafloor cabled observation system, operated by NIED, has considerably contributed to the detection of the shallow tremor along the Japan Trench.^63),64) The properties of the shallow VLFEs are similar to those of deep tremor because shallow tremor is associated with shallow VLFEs and indicates clear along-strike migration.

3-2-3. Shallow SSEs.

The ocean-bottom pressure gauges detected two shallow SSEs with respect to the rupture area of the Tohoku earthquake in 2008 and 2011, before its occurrence.⁶⁵⁾ Furthermore, two foreshock sequences that occurred one month and two days before the 2011 Tohoku earthquake were associated with migration toward the initial rupture area of the Tohoku earthquake.⁶⁶⁾ These two foreshock sequences included repeating earthquakes, such that the shallow SSEs occurred as a background foreshock process. The repeating migration sequences of the SSE slip front may have resulted in additional stress with respect to the initial rupture area of the Tohoku earthquake in 2011, triggering the great earthquake rupture.

Borehole pore-pressure gauges were installed offshore to the southeast of the Kii Peninsula, and detected the pressure changes associated with the occurrence of shallow VLFEs and tremor observed by DONET.⁶⁷⁾ These pressure changes indicate shallow SSEs with reverse faulting near the Nankai Trench axis. Furthermore, the pressure-change time series is consistent with the cumulative seismic moment of shallow VLFEs, indicating that shallow SSEs and VLFEs can be attributed to a common slip process on the same fault plane.⁵⁷⁾

The heterogeneous distribution of interplate coupling in the offshore region of the Nankai Trough was estimated based on the results of a widely deployed seafloor acoustic GNSS network.⁶⁸⁾ Furthermore, transient seafloor displacement was observed near the Kii Channel region at a couple of acoustic GNSS stations from 2018 to 2020, and the SSE fault plane was estimated at the Nankai Trough.⁶⁹⁾ A similar long-term shallow SSE was observed in 2009 in the same region.

SSEs with an approximate M_W of 6.7 were initially detected at Iriomote Island in the Ryukyu Islands between Kyushu and Taiwan.⁷⁰⁾ These SSEs usually last for approximately one month and have an interval of six months. Furthermore, many short-term SSEs with an approximate M_W of 6.4 and a duration of several days have been detected near the Okinawa and other nearby islands.⁷¹⁾ Many SSEs in the Ryukyu Islands are classified as deep slow earthquakes; however, some SSEs may be classified as shallow slow earthquakes because they are located near the Ryukyu Trench and denote interactions with shallow VLFEs.

4. Characteristic spatiotemporal activity and interaction of slow earthquakes

Each type of slow earthquake in the Nankai subduction zone is mainly distributed horizontally along the strike of the subducting plate geometry at nearly the same depth and is usually segmented. These slow earthquakes are characterized by along-dip systematic changes and along-strike heterogeneities. Thus, each phenomenon is controlled by a unique frictional property at the specified geophysical and/or geological condition. The slow earthquakes in the Nankai subduction zone are mainly separated into shallow and deep phenomena relative to the depth of the seismogenic zone of great earthquakes. This bimodal depth distribution is consistent with the depth-dependent frictional properties along the subducting plate interface,⁵⁾ corresponding to the existence of a conditionally stable field representing the transition between unstable and stable regions in the shallow and deep sections of the seismogenic zone. In this section, the characteristic properties of slow earthquakes, especially those in the Nankai subduction zone, are reviewed by considering the spatiotemporal interaction between slow earthquakes.

4-1. Along-strike heterogeneity of deep slow earthquakes.

4-1-1. Segmentation of deep ETS.

ETS, which is the deepest interplate slow earthquake phenomenon been recognized to date and is well observed in the Nankai and Cascadia subduction zones, shows an along-strike variation roughly characterized by two different scale lengths: segment scale and subsegment scale.

The spatiotemporal properties of ETS are characterized by its segmentation and periodicity even though the scales vary in both space and time. The along-strike extensions of the ETS zone are 600 and 1,200 km in Nankai and Cascadia, respectively. However, the entire ETS source region is mainly divided into segments with different recurrence intervals similarly for both subduction zones. Each segment group includes some small segments, wherein ETS occurs at a similar recurrence interval but with different timing. There are three main segment groups in Cascadia, with 14-, 19-, and 11-month recurrence intervals in Vancouver Island and northern Washington, southern Washington and Oregon, and northern California, respectively.⁷²⁾ The ETS in southwest Japan is also divided into three segment groups, with 6-, 3-, and 6-month recurrence intervals in western Shikoku, eastern Shikoku and western Kii Peninsula, and northeastern Kii Peninsula and Tokai, respectively.¹⁸⁾

The ETS behavior may be controlled by the hanging wall structure. Brudzinski and Allen⁷²⁾ proposed that the three segment groups of ETS in Cascadia correspond to different geologic terranes. The recurrence interval and elevation of the large-scale upper plate topography are negatively correlated. However, there is no such relation between the large-scale topography and segmentation of ETS in southwest Japan. Audet and Burgmann⁷³⁾ recently assembled a global compilation of the Vp/Vs ratio with respect to the hanging wall just above the ETS zone and demonstrated that the ETS recurrence interval was proportional to Vp/Vs. Nakajima and Hasegawa⁷⁴⁾ investigated the relation between the tremor location and the three-dimensional seismic velocity structure around the subducting Philippine Sea plate boundary in southwest Japan. They revealed that the P-wave velocity at the bottom of the upper plate was normal just above the tremor region; however, it was low above the tremorless region. This low-velocity body on the hanging wall just above the tremorless region is interpreted as the metamorphism related to fluid enrichment from the subducting slab. Tremor actively occurs just beneath the unmetamorphosed overriding plate because of the impermeable plate interface and enhanced pore fluid pressure.

The segment boundary of ETS illuminated by the multisegment rupture of ETS may be affected by the spatial frictional heterogeneity similar to the relation between the along-strike variation of the fault friction and earthquake faulting.⁷⁵⁾ Tremor migration usually occurs within each segment; however, this migration rarely extends to the neighboring segments. The ETS migration pattern is unique for each segment.¹⁸⁾ The tremor episode usually begins from the edge of the tremor gap at Ise Bay and propagates southwestward and eastward in Kii and Tokai, respectively. Such a tremor gap is generally used to define the segment boundary. There are three tremor gaps in southwest Japan, i.e., Ise Bay, Kii Channel, and between the eastern and middle parts of Shikoku. These tremor gaps are always seismically quiet; however, tremor migration can continue smoothly along both sides of a gap. For example, a 2006 tremor episode began in central Kii Peninsula and migrated northeastward, ending at Ise Bay, whereas another tremor episode propagated eastward through the Tokai segment. Both the tremor migration pathways are smoothly connected by a single regression line. These migrating tremor episodes are usually associated with short-term SSEs. Generally, the SSE source faults are estimated based on the spatiotemporal distribution of tremor. However, the SSE source faults can also be estimated in the aseismic portion at Ise Bay,^27),76) indicating that the tremor patch was not homogeneously distributed on the SSE fault plane, with no or few tremor patches at Ise Bay. The smooth migration of tremor can be frequently observed along both sides of the tremor gap between eastern and middle Shikoku. There is no geodetic evidence to prove the occurrence of short-term SSEs in the gap; however, SSEs may propagate smoothly along the plate interface, and the tremor gap can be attributed to the absence of a tremor patch on the plate interface.⁷⁷⁾

4-1-2. Subsegment-scale heterogeneity of ETS.

The ETS comprising tremor, VLFEs, and short-term SSEs does not occur uniformly within each segment; however, it indicates an along-strike variation, which can possibly denote the spatial heterogeneity in the plate boundary zone. The tremor distribution is rather homogeneous even though a heterogeneous distribution of VLFEs and short-term SSE faults can be observed in some areas.¹⁸⁾ Similarly, improvements of the tremor detection method, including the estimation of the radiation energy, provide a more heterogeneous distribution similar to that observed in case of VLFEs.⁷⁸⁾ The along-strike profiles of the tremor radiation energy and VLFE moment are similar and consistent with the total slip distribution of the short-term SSEs estimated via GNSS analysis,³⁴⁾ including some discrepancies in the relation between SSEs and seismic slow earthquakes. The general consistency between SSEs and seismic slow earthquakes supports an interpretation for the occurrence of seismic slow earthquakes modulated by aseismic slow slips.³⁴⁾

The clear difference in the tremor activity pattern in the western Shikoku region may suggest the along-strike heterogeneity of the tremor patch strength⁷⁹⁾ based on the energy information of the tremor catalog.⁷⁸⁾ The radiation energy is larger and the occurrence number is smaller than those in the inland region for the offshore Bungo Channel region. Furthermore, many tremor events have been initiated in the offshore Bungo Channel region that has diffusively migrated eastward. The tremor migration velocity is proportional to the radiation energy. The aforementioned observational results are explained using a model in which the tremor patch associated with the spatially heterogeneous strength distribution ruptures in a chain via the stress diffusion process.⁸⁰⁾ This model indicates that strong and weak tremor patches are distributed in the offshore and inland regions, respectively.

4-1-3. Segmentation of long-term SSEs.

Long-term SSEs, which exhibit considerably longer durations and recurrence intervals than short-term SSEs, occur between the ETS and seismogenic zones of great earthquakes in the Nankai subduction zone. The along-strike distribution of long-term SSEs is segmented similar to ETS, although the distribution pattern is different. ETS is distributed from the southern Nagano Prefecture to the Bungo Channel and includes some gaps; however, the long-term SSE distribution is considerably heterogeneous, and isolated occurrences at Lake Hamana and continuous occurrences from the Kii Channel to the southern Miyazaki Prefecture can be observed along the depth contours of the subducting Philippine Sea plate. Until now, long-term SSEs have not been detected in the Kii Peninsula region, which may denote the lack of spatial gap between the ETS distribution and the seismogenic zone in this region. However, there is no ETS activity downdip of the long-term SSEs in eastern Kyushu, although both ETS and long-term SSEs occur in Shikoku. These variations in slow earthquake activity may be affected by the steep geometry of the subducting oceanic plate located beneath Kyushu. Even though the geometry of the ETS source region differs, long-term SSEs frequently exhibit along-strike migration, as observed in the ETS activity, from southern Kyushu to central Shikoku.⁴³⁾ The long-term SSE source regions are separated into four segments based on the distribution of the total slip amounts and recurrence interval, and a different slip amount can be observed along each segment. Further, a small slip was observed on the segment that is a downward extension of the coseismic rupture areas of the previous great earthquakes, whereas a large slip was observed on the segment that is a downward extension of the areas with no coseismic rupture. Thus, the long-term SSE activity is controlled by the status of updip coupling. Furthermore, repeating earthquake analysis has demonstrated that the northward migration of long-term SSEs in eastern Kyushu is correlated with the slip acceleration in the quasi-stable slip region updip of the long-term SSE source faults because the repeating earthquakes occur on seismic patches surrounded by quasi-stable slip regions.⁸¹⁾ Therefore, the long-term SSE activity in the southeastern Kyushu region strongly interacts with the quasi-stable sliding updip of the long-term SSE source faults. Furthermore, there are large variations in along-strike coupling, including a quasi-stable sliding portion because of repeaters and large earthquake coseismic source regions.

4-2. Along-dip variation.

4-2-1. Depth-dependent recurrence behavior of ETS.

Based on the depth-dependent frictional properties, the ETS recurrence interval is observed to vary systematically in the along-dip direction locations within each segment. The ETS recurrence interval in southwest Japan and Cascadia indicates similar depth dependence.^82),83) Minor ETS episodes can be frequently observed along the downdip edge of the ETS zone, whereas major ETS episodes can be infrequently observed along the updip edge of the ETS zone. The recurrence interval systematically decreased with the increasing along-dip depth. Major ETS episodes exhibit a unique recurrence interval for each segment. They are usually initiated as a minor event from the downdip edge of the ETS zone, migrate updip during the early stage, and finally develop into an active episode associated with along-strike propagation. Such a depth-dependent recurrence behavior is explained by the upward stress transfer model proposed by Wech and Creager.⁸³⁾ The frictional strength is expected to become weaker with increasing depth because of the systematic change in temperature and/or pore fluid pressure. Minor ETS episodes can be readily observed at the downdip edge of the ETS zone because of stress accumulation at the boundary between ETS and stable sliding zones. The frequent occurrence of minor ETS events is initiated in the deep section. Subsequently, the loading of stress onto the shallower section is gradually initiated, and propagation of major ETS episodes to the updip edge can be observed.

4-2-2. Interaction between ETS and long-term SSEs.

The long-term SSEs occurring at the updip part from ETS may trigger ETS tremor activity. This triggering effect is considerably dependent on the distance between these slow earthquake phenomena. Obvious interactions can be observed between the long-term SSEs and ETS in the Bungo Channel.⁸⁴⁾ The tremor activity along the updip section of the tremor zone, which is the adjacent downdip section of the long-term SSE source fault, is only activated when long-term SSEs occur in the Bungo Channel; however, the tremor activity along the downdip section of the tremor zone is stable and unrelated to the occurrence of long-term SSEs. If the magnitudes of the slow earthquake phenomenon interacting with each other are compared, then the long-term SSE and tremor must be the master and slave events, respectively. The activation of the tremor along the updip part of the tremor zone does not increase smoothly but comprises many step-like increases, although the GNSS displacement changes that represent the occurrence of the long-term SSEs are smooth. Each small step indicates a small tremor burst, suggesting the occurrence of a minor short-term SSE not detected geodetically. The short-term SSEs are triggered to reach the slip caused by long-term SSEs under the displacement continuum condition at the boundary section.

The two small long-term SSEs in the Bungo Channel in 2014 and 2016 clearly show a distance-dependent interaction during the 2010–2019 period when there were no large long-term SSEs. These small long-term SSEs exhibit slightly different slip areas. The SSEs during 2014 and 2016 are located in the northern and southern parts of the Bungo Channel, respectively, and the differences in their slip locations indicate considerably different tremor activities. An obvious tremor activity was observed during the long-term SSE in 2014, whereas no tremor activity was associated with the long-term SSE observed in 2016. The difference in tremor behavior can also be explained by the difference in the distances between the master and slave phenomena. The long-term SSE observed in 2014 was located in the northern section close to the tremor zone, whereas that in 2016 was located far from the tremor zone. Deep VLFEs were triggered by the Bungo Channel long-term SSEs. Baba et al.³¹⁾ detected deep VLFEs in western Shikoku using the matched filter method and revealed a rapid increase in VLFE seismicity with respect to the long-term SSEs observed in 2010 and 2014. The number and activated area of the triggered VLFEs during the SSE in 2010 were larger than those of the SSE in 2014 because of the larger moment of the former.

The interaction between long-term SSEs and tremor has also been observed in Tokai and Mexico,⁸⁵⁾ although the interactions are not so clear. Further, the interactions are considerably affected by the distance of the master long-term SSE from the slave tremor. The slip velocity of the master event may affect the degree of the triggered phenomenon. Conversely, annual-scale interactions between tremor activity and long-term SSEs have been observed in western and central Shikoku.⁴¹⁾ The eastward migration of long-term SSEs has been observed twice after the Bungo Channel long-term SSEs in 2003 and 2010 along the gap zone between the ETS and seismogenic zones. The tremor activity pattern migrates eastward at 50 km/year, which is similar to the eastward migration of long-term SSEs.

4-2-3. Controlling the difference between ETS and long-term SSEs based on the structural environment.

Different frictional regimes that can be attributed to different geological materials may result in slow earthquakes with different slip velocities. Long-term SSEs and ETS (short-term SSEs) located next to each other in the Bungo Channel may be controlled by different hanging wall materials, i.e., the mantle wedge and continental lower crust, respectively. Kato et al.⁸⁶⁾ investigated a high-resolution subsurface structure in the Tokai region via tomography and receiver function analyses of the seismic data obtained from a temporally deployed linear array above both the ETS and long-term SSE source region. Further, they concluded that the long-term SSEs and ETS were separated by the continental Moho discontinuity; thus, the long-term SSEs and ETS were located at the base of the continental lower crust and the forearc mantle wedge, respectively. Holtkamp and Brudzinski⁸⁷⁾ estimated the interplate coupling along the subducting plate interface in Cascadia based on the GPS data and revealed that the ETS region at the base of the mantle wedge was considerably coupled, whereas the region beneath the continental lower crust was weakly coupled because of the geological material, which is governed by the flow law. The interplate slip properties in the zone between the ETS and locked zones may be affected by the viscoelastic nature of the lower crust even though long-term SSEs have never been detected in Cascadia.

4-2-4. Interaction between the long-term SSE and shallow VLFEs.

In the Bungo Channel, long-term SSEs may interact with the shallow VLFEs near the Nankai Trough because of a possible SSE between them near the western margin of the large rupture area owing to the Nankai earthquake. Hirose et al.⁸⁴⁾ observed that the long-term SSEs in the Bungo Channel interacted with the deep tremor activity along the updip section of the ETS zone and shallow VLFEs near the Nankai Trough in 2003 and 2010. The correlation pattern slightly differed in case of deep tremor and shallow VLFEs. The activated deep tremor continued for several months corresponding to the duration of long-term SSEs; however, the shallow VLFEs were activated for several days, which coincided with the beginning of the GPS displacement changes that can be attributed to long-term SSEs. Although the detailed triggering mechanisms may differ, both the deep tremor and shallow VLFEs were likely triggered by long-term SSEs because of the reproducibility of coherent observations in 2003 and 2010. However, the source location of the activated shallow VLFEs was >100 km away from the estimated source fault of the long-term SSEs. Both the source locations were estimated based on the inland observation, although the poor resolution of the inland GPS network limited the estimation of the offshore slip distribution. The relation between the long-term SSEs and deep tremor suggested that shallow VLFEs may be triggered by the slow slip occurring near the shallow VLFE source region. The activation of shallow VLFEs would be explained if the long-term SSE source fault extended southward near the trough axis.⁸⁴⁾ If such an extended slip distribution exists for long-term SSEs, then different types of slow earthquakes should align in the along-dip direction; further, the alignment of slow earthquakes corresponds to the western boundary of the coseismic rupture associated with the Nankai earthquake that occurred in 1946. Such a slow earthquake alignment may act as a barrier to high-speed coseismic rupture.⁸⁴⁾

Uchida et al.⁸¹⁾ reinvestigated the relation between shallow VLFEs, deep tremor, and long-term SSEs in addition to the repeating earthquake activity. They reported the northward migration of the slow earthquake activity from the southern extent of the shallow VLFE source region in Hyuga-nada to the deep tremor region in western Shikoku through the updip section of the repeating earthquake region and long-term SSE fault areas in the Bungo Channel. Although this denotes an opposite migration pattern when compared with that proposed by Hirose et al.,⁸⁴⁾ various types of slow earthquakes may mutually interact with each other over a wide region from the shallow to deep sections in the Hyuga-nada and Bungo Channel regions.

Nakamura and Sunagawa⁸⁸⁾ revealed an interaction between shallow VLFEs and repeating long-term SSEs in the southwestern Ryukyu Trench at a depth of 30–50 km beneath the southern Ryukyu arc. The VLFE seismicity peaked 10–20 days after the onset of the long-term SSEs.

4-3. Interactions between shallow slow earthquakes.

The plate interface at shallow depths near the trench axis is considerably heterogeneous, including a strongly coupled section containing coseismic ruptures because of great earthquakes and a weakly coupled section in which slow earthquakes occurred. The acoustic GNSS observations have revealed a contrast in coupling strength along the Nankai Trough axis and indicated that shallow VLFEs are distributed across the weakly coupled section.⁶⁸⁾ A shallow tremor was not detected within the large coseismic rupture area of the Tohoku earthquake in 2011 along the Japan Trench.⁶³⁾ How do these seismic slow earthquakes along the shallow section interact with the geodetic slow earthquakes? The short-term SSEs detected by a borehole pore-pressure gauge were clearly correlated temporally with the shallow VLFE and shallow tremor seismicity detected by DONET offshore to the southeast of the Kii Peninsula.^57),67)

Conversely, the shallow VLFE seismicity in the Tokachi offshore region was very coherent with the crustal deformation caused by the afterslip from the Tokachi-oki earthquake in 2003.⁵³⁾ Similarly, the shallow VLFE seismicity was activated because of the afterslip sequence in the northern and southern sections of the large coseismic rupture zone of the Tohoku earthquake in 2011.⁵⁹⁾ These shallow VLFEs were activated by a transient slip with a long characteristic timescale. However, only some long-term SSE observations are available. Long-term SSEs with a duration in the range of years may be detected in 2009 and 2018 in the Kii Channel offshore region near the Nankai Trough axis.⁶⁹⁾ These long-term SSEs have been associated with the migrating shallow VLFE episodes.⁸⁹⁾

5. Discussion

The slow earthquake activity in southwest Japan is roughly characterized by a zone-like distribution associated with the along-strike heterogeneity at different depth ranges, although the slip behaviors may vary in two dimensions on the subducting plate interface (Fig. 5). Long-term SSEs were initially detected in only two regions, Tokai and Bungo Channel in the early 2000s, and the gap between the ETS and locked zones was filled by long-term SSEs. Therefore, the long-term SSEs and tremor systematically share the megathrust plate boundary at each depth and are horizontally distributed, including their separation into small segments. This suggests that different types of slow earthquakes occur along the deep section of the Nankai subduction zone. However, each type of slow earthquake has a different along-strike extension. The long-term SSEs are distributed beneath both Kyushu and Shikoku, although there is no ETS activity beneath Kyushu, except the minor and localized tremor activity in southeastern Kyushu.⁹⁰⁾

Fig. 5.

Schematic of the slow earthquake activity in the Nankai Trough region (left) and Japan Trench region (right). The yellow ellipses indicate deep ETS and shallow slow earthquakes, which are characterized by short-term SSEs with durations of days to weeks. The green ellipses indicate long-term SSEs with durations of months to years. The red ellipses indicate the locked zones, which are considered to be the seismogenic zones for the M_W 7–9 earthquakes. The blue ellipses and small red circles indicate quasi-stable sliding zones with many repeating earthquakes (REs).

Conversely, shallow tremor occurred along the Japan Trench in the offshore region of northeast Japan at the same depth as that of the great earthquake coseismic rupture area along the subducting Pacific plate boundary.⁶³⁾ The shallow slow earthquake activity along the Japan Trench exhibited no systematic pattern but indicated strong along-strike heterogeneity, including the fast-speed coseismic rupture of regular earthquakes. Such along-strike heterogeneity was observed in the shallow VLFE distribution⁹¹⁾ and interplate coupling distribution⁶⁸⁾ along the Nankai Trough.

Slow earthquakes occur on the updip (shallower) and downdip (deeper) sections of the seismogenic zone in southwest Japan. The ETS activity and a couple of different tremor migration modes are commonly observed in both the depth sections even though the pressure and temperature environments are considerably different. However, the detailed activity pattern of slow earthquakes is slightly different. For example, the occurrence of slow earthquakes is periodic in the deeper section compared with that in the shallower section. The shallow tremor migration velocities were 20 km/day and 100 km/day for the two migration modes observed in Hyuga-nada,⁸⁾ which were a few times higher than the deep tremor migration velocities. Such minor differences in activity patterns may be useful for resolving the frictional properties via experimental studies. The observations of various slow earthquakes in the offshore area should be considered to resolve these properties.

The slow earthquakes and interplate great earthquakes are related since the initial discovery of deep tremor in the Nankai subduction zone because of their neighboring spatial geometries. Obara and Kato³⁷⁾ proposed three key aspects with respect to the connection between slow and great earthquakes. The slow earthquakes are initially considered to be an analog for great earthquakes because of their similar activity patterns. Both the slow and great earthquakes are characterized by a periodic recurrence behavior, and large fluctuations can be observed in each segment (Fig. 6).^39),92),93) Therefore, frequent slow earthquake observations will be useful to obtain the occurrence model of great earthquakes based on their large recurrence interval. A slow earthquake is subsequently considered to be a stress meter for monitoring the stress changes in the locked section because of its sensitive nature, as suggested from triggering by teleseismic waves and tidal modulation. Although there is no example to indicate that the slow earthquake activity changes as a function of stress based on the accumulated stress in the locked section in southwest Japan, there are a couple of examples demonstrating changes in slow earthquake activity prior to the occurrence of historical large earthquakes around the world.^94)–96) The occurrence of slow earthquakes is also expected to transfer the stress to the surrounding region, including the locked portion. The remainder of this section will discuss stress transfer based on the recent research results.

Fig. 6.

Recurrence patterns for different slip phenomena. (Top) Great earthquakes in the Nankai subduction zone.⁹²⁾ Vertical bars indicate the along-strike rupture area. The red lines, orange dashed lines, and yellow dashed lines are classified as more reliable to less reliable results. The number between bars indicate the years of interval between great earthquakes. (Middle) Long-term SSEs.³⁹⁾ The blue bars denoted by T indicate the long-term SSEs in Tokai. The green bar denoted by KC indicates the long-term SSE in the Kii Channel. The brown bar denoted by S1 and green bar denoted by S2 indicate the long-term SSEs in central Shikoku. The red bars denoted by B indicate the long-term SSEs in the Bungo Channel. The purple bars denoted by H indicate the long-term SSEs in Hyuga-nada. The gray bars indicate the tremor source regions. (Bottom) ETS.⁹³⁾ The red dots indicate tremor.

The migration of slow earthquakes toward the initial rupture of the Tohoku earthquake in 2011⁶⁶⁾ is a typical example of the stress transfer phenomenon. A similar slow earthquake migration sequence was observed before the Iquique earthquake in 2014 in northern Chile.⁹⁷⁾ Although the amount of stress transferred from a given slow earthquake sequence is likely to be considerably small for triggering the rupture of a great earthquake, the frequent occurrence of slow earthquakes may result in stress accumulation, gradually approaching the critical rupture level. Takagi et al.⁴³⁾ revealed the northward migration of long-term SSEs along the subducting Philippine Sea plate, where the long-term SSEs migrated from the weakly coupled region in southeastern Kyushu to the firmly locked zone beneath Shikoku. Thus, a long-term SSE may load additional stress to the locked zone over a characteristic timescale of years. The ETS activity along the deepest section of the transition zone indicates upward stress transfer within the ETS zone, as previously discussed, which may cause stress accumulation at the updip section.⁸³⁾ If the plate boundary toward the updip side of the ETS zone is in contact with the continental lower crust in the hanging wall, then the updip stress transfer may relax because of the viscous properties of the hanging wall. However, if the ETS zone is in direct contact with the locked zone, then the updip stress loading may result in the accumulation of additional stress in the locked zone. Kano et al.⁹⁸⁾ revealed that the small slip along the downdip portion of the locked zone was associated with the ETS episodes in western Shikoku. They interpreted that the ETS episode ruptured the impermeable seal at the top of the mantle wedge such that the fluid pressure could propagate upward along the plate interface to the downdip portion of the locked zone through the base of the continental lower crust, which is the source region of long-term SSEs. Thus, ETS may enable stress transfer to the locked section in both the cases even if the continental lower crust does not exist in the hanging wall.

6. Summary

The slow earthquakes along the subducting plate interface are classified into various seismic and geodetic phenomena exhibiting different characteristic timescales at different depths. The segmentation and large fluctuations in recurrence intervals are common properties associated with the slow earthquakes detected at the downdip edge of the seismogenic zone; however, shallow slow earthquakes also occur near the locked zones containing high-speed coseismic rupture zones. The frequently observed interaction between different types of slow earthquakes may extend to the neighboring seismogenic zone. The stress accumulation in the locked section may change the activity patterns of the neighboring slow earthquakes. Therefore, precisely monitoring slow earthquakes is important for evaluating great earthquakes in future, which is dependent on the long-term maintenance and continued improvement of the high-quality observation networks.

Acknowledgments

The author would like to thank the anonymous reviewers for their valuable comments and kind suggestions to improve the manuscript. The author would also like to thank J.P. Vilotte from the Institute de Physique du Globe de Paris and K.C. Creager from the University of Washington for providing overseas research opportunities, including writing this manuscript. This research was supported by JSPS KAKENHI Grant# JP16H06473 in Grant-in-Aid for Scientific Research on Innovative Areas “Science of Slow Earthquakes” and Grant projects for short-term overseas research program supported by Kajima Foundation for academy promotion.

Notes

Edited by Yoshio FUKAO, M.J.A.

Correspondence should be addressed: K. Obara, Earthquake Research Institute, The University of Tokyo, 1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-0032, Japan (e-mail: [e-mail]).

References

• 1)   Hirose,  H.,  Hirahara,  K.,  Kimata,  F.,  Fujii,  N. and  Miyazaki,  S. (1999) A slow thrust slip event following the two 1996 Hyuganada earthquakes beneath the Bungo Channel, southwest Japan. Geophys. Res. Lett. 26, 3237–3240.
• 2)   Obara,  K. (2002) Nonvolcanic deep tremor associated with subduction in southwest Japan. Science 296, 1679–1681.
• 3)   Nadeau,  R.M. and  Dolenc,  D. (2005) Nonvolcanic tremors deep beneath the San Andreas Fault. Science 307, 389.
• 4)   Peng,  Z. and  Chao,  K. (2008) Non-volcanic tremor beneath the Central Range in Taiwan triggered by the 2001 M_W 7.8 Kunlun earthquake. Geophys. J. Int. 175, 825–829.
• 5)   Scholz,  C. (1998) Earthquakes and friction laws. Nature 391, 37–42.
• 6)   Hyndman,  R.,  Wang,  K. and  Yamano,  M. (1995) Thermal constraints on the seismogenic portion of the southwestern Japan subduction thrust. J. Geophys. Res. Solid Earth 100, 15373–15392.
• 7)   Obara,  K. and  Ito,  Y. (2005) Very low frequency earthquake excited by the 2004 off the Kii peninsula earthquake: A dynamic deformation process in the large accretionary prism. Earth Planets Space 57, 321–326.
• 8)   Yamashita,  Y.,  Yakiwara,  H.,  Asano,  Y.,  Shimizu,  H.,  Uchida,  K.,  Hirano,  S. (2015) Migrating tremor off southern Kyushu as evidence for slow slip of a shallow subduction interface. Science 348, 676–679.
• 9)   Okada,  Y.,  Kasahara,  K.,  Hori,  S.,  Obara,  K.,  Sekiguchi,  S.,  Fujiwara,  H. (2004) Recent progress of seismic observation networks in Japan—Hi-net, F-net, K-net and KiK-net. Earth Planets Space 56, xv–xxviii.
• 10)   Sagiya,  T. (2004) A decade of GEONET: 1994–2003—The continuous GPS observation in Japan and its impact on earthquake studies. Earth Planets Space 56, xxix–xli.
• 11)   Obara,  K.,  Kasahara,  K.,  Hori,  S. and  Okada,  Y. (2005) A densely distributed high-sensitivity seismograph network in Japan: Hi-net by National Research Institute for Earth Science and Disaster Prevention. Rev. Sci. Instrum. 76, 021301.
• 12)   Müller,  R.D.,  Sdrolias,  M.,  Gaina,  C. and  Roest,  W.R. (2008) Age, spreading rates, and spreading asymmetry of the world’s ocean crust. Geochem. Geophys. Geosyst. 9, Q04006.
• 13)   Rogers,  G. and  Dragert,  H. (2003) Episodic tremor and slip on the Cascadia subduction zone: The chatter of silent slip. Science 300, 1942–1943.
• 14)   Shelly,  D.R.,  Beroza,  G.C. and  Ide,  S. (2007) Non-volcanic tremor and low-frequency earthquake swarms. Nature 446, 305–307.
• 15)   Maeda,  T. and  Obara,  K. (2009) Spatiotemporal distribution of seismic energy radiation from low-frequency tremor in western Shikoku, Japan. J. Geophys. Res. Solid Earth 114, B00A09.
• 16)   Ide,  S. (2010) Striations, duration, migration and tidal response in deep tremor. Nature 466, 356–360.
• 17)   Katsumata,  A. and  Kamaya,  N. (2003) Low-frequency continuous tremor around the Moho discontinuity away from volcanoes in the southwest Japan. Geophys. Res. Lett. 30, 1020.
• 18)   Obara,  K. (2010) Phenomenology of deep slow earthquake family in southwest Japan: Spatiotemporal characteristics and segmentation. J. Geophys. Res. Solid Earth 115, B00A25.
• 19)   Obara,  K.,  Hirose,  H.,  Yamamizu,  F. and  Kasahara,  K. (2004) Episodic slow slip events accompanied by non-volcanic tremors in southwest Japan subduction zone. Geophys. Res. Lett. 31, L23602.
• 20)   Ide,  S.,  Shelly,  D.R. and  Beroza,  G.C. (2007) Mechanism of deep low frequency earthquakes: Further evidence that deep non-volcanic tremor is generated by shear slip on the plate interface. Geophys. Res. Lett. 34, L03308.
• 21)   Houston,  H.,  Delbridge,  B.G.,  Wech,  A.G. and  Creager,  K.C. (2011) Rapid tremor reversals in Cascadia generated by a weakened plate interface. Nat. Geosci. 4, 404–409.
• 22)   Obara,  K.,  Matsuzawa,  T.,  Tanaka,  S. and  Maeda,  T. (2012) Depth-dependent mode of tremor migration beneath Kii Peninsula, Nankai subduction zone. Geophys. Res. Lett. 39, L10308.
• 23)   Shelly,  D.R.,  Beroza,  G.C. and  Ide,  S. (2007) Complex evolution of transient slip derived from precise tremor locations in western Shikoku, Japan. Geochem. Geophys. Geosyst. 8, Q10014.
• 24)   Miyazawa,  M. and  Mori,  J. (2006) Evidence suggesting fluid flow beneath Japan due to periodic seismic triggering from the 2004 Sumatra–Andaman earthquake. Geophys. Res. Lett. 33, L05303.
• 25)   Obara,  K. (2003) Time sequence of deep low frequency tremors in the southwest Japan subduction zone: Triggering phenomena and periodic activity. J. Geog. 112, 837–849 (in Japanese with English abstract).
• 26)   Nakata,  R.,  Suda,  N. and  Tsuruoka,  H. (2008) Non-volcanic tremor resulting from the combined effect of Earth tides and slow slip events. Nat. Geosci. 1, 676–678.
• 27)   Ito,  Y.,  Obara,  K.,  Shiomi,  K.,  Sekine,  S. and  Hirose,  H. (2007) Slow earthquakes coincident with episodic tremors and slow slip events. Science 315, 503–506.
• 28)   Tsuruoka,  H.,  Kawakatsu,  H. and  Urabe,  T. (2009) GRiD MT (Grid-based Realtime Determination of Moment Tensors) monitoring the long-period seismic wavefield. Phys. Earth Planet. 175, 8–16.
• 29)   Ito,  Y.,  Obara,  K.,  Matsuzawa,  T. and  Maeda,  T. (2009) Very low frequency earthquakes related to small asperities on the plate boundary interface at the locked to aseismic transition. J. Geophys. Res. Solid Earth 114, B00A13.
• 30)   Ide,  S. and  Yabe,  S. (2014) Universality of slow earthquakes in the very low frequency band. Geophys. Res. Lett. 41, 2786–2793.
• 31)   Baba,  S.,  Takeo,  A.,  Obara,  K.,  Kato,  A.,  Maeda,  T. and  Matsuzawa,  T. (2018) Temporal activity modulation of deep very low frequency earthquakes in Shikoku, Southwest Japan. Geophys. Res. Lett. 45, 733–738.
• 32)   Dragert,  H.,  Wang,  K. and  James,  T.S. (2001) A silent slip event on the deeper Cascadia subduction interface. Science 292, 1525–1528.
• 33)   Hirose,  H. and  Obara,  K. (2005) Repeating short- and long-term slow slip events with deep tremor activity around the Bungo channel region, southwest Japan. Earth Planets Space 57, 961–972.
• 34)   Nishimura,  T.,  Matsuzawa,  T. and  Obara,  K. (2013) Detection of short-term slow slip events along the Nankai Trough, southwest Japan, using GNSS data. J. Geophys. Res. Solid Earth 118, 3112–3125.
• 35)   Hirose,  H. and  Obara,  K. (2010) Recurrence behavior of short-term slow slip and correlated nonvolcanic tremor episodes in western Shikoku, southwest Japan. J. Geophys. Res. Solid Earth 115, B00A21.
• 36)   Bartlow,  N.M.,  Miyazaki,  S.,  Bradley,  A.M. and  Segall,  P. (2011) Space-time correlation of slip and tremor during the 2009 Cascadia slow slip event. Geophys. Res. Lett. 38, L18309.
• 37)   Obara,  K. and  Kato,  A. (2016) Connecting slow earthquakes to huge earthquakes. Science 353, 253–257.
• 38)   Ozawa,  S.,  Murakami,  M.,  Kaidzu,  M.,  Tada,  T.,  Hatanaka,  Y.,  Yarai,  H. (2002) Detection and monitoring of ongoing aseismic slip in the Tokai region, Central Japan. Science 298, 1009–1012.
• 39)   Kobayashi,  A. (2014) A long-term slow slip event from 1996 to 1997 in the Kii Channel, Japan. Earth Planets Space 66, 9.
• 40)   Kobayashi,  A. (2017) Objective detection of long-term slow slip events along the Nankai Trough using GNSS data (1996–2016). Earth Planets Space 69, 171.
• 41)   Takagi,  R.,  Obara,  K. and  Maeda,  T. (2016) Slow slip event within a gap between tremor and locked zones in the Nankai subduction zone. Geophys. Res. Lett. 43, 1066–1074.
• 42)   Ozawa,  S. (2017) Long-term slow slip events along the Nankai trough subduction zone after the 2011 Tohoku earthquake in Japan. Earth Planets Space 69, 56.
• 43)   Takagi,  R.,  Uchida,  N. and  Obara,  K. (2019) Along-strike variation and migration of long-term slow slip events in the western Nankai subduction zone, Japan. J. Geophys. Res. Solid Earth 124, 3853–3880.
• 44)   Yarai,  H. and  Ozawa,  S. (2013) Quasi-periodic slow slip events in the afterslip area of the 1996 Hyuga-nada earthquakes, Japan. J. Geophys. Res. Solid Earth 118, 2512–2527.
• 45)   Miyazaki,  S.,  Segall,  P.,  McGuire,  J.J.,  Kato,  T. and  Hatanaka,  Y. (2006) Spatial and temporal evolution of stress and slip rate during the 2000 Tokai slow earthquake. J. Geophys. Res. Solid Earth 111, B03409.
• 46)   Yamamoto,  E.,  Matsumura,  S. and  Ohkubo,  T. (2005) A slow slip event in the Tokai area detected by tilt and seismic observation and its possible recurrence. Earth Planets Space 57, 917–923.
• 47)   Ozawa,  S.,  Tobita,  M. and  Yarai,  H. (2016) A possible restart of an interplate slow slip adjacent to the Tokai seismic gap in Japan. Earth Planets Space 68, 54.
• 48)   Hirose,  H.,  Kimura,  H.,  Enescu,  B. and  Aoi,  S. (2012) Recurrent slow slip event likely hastened by the 2011 Tohoku earthquake. Proc. Natl. Acad. Sci. U.S.A. 109, 15157–15161.
• 49)   Kato,  A.,  Igarashi,  T. and  Obara,  K. (2014) Detection of a hidden Boso slow slip event immediately after the 2011 M_W 9.0 Tohoku-Oki earthquake, Japan. Geophys. Res. Lett. 41, 5868–5874.
• 50)   Fukuda,  J. (2018) Variability of the space-time evolution of slow slip events off the Boso Peninsula, central Japan, from 1996 to 2014. J. Geophys. Res. Solid Earth 123, 732–760.
• 51)  Ishihara, Y. (2003) Major existence of very low frequency earthquakes in background seismicity along subduction zone of south-western Japan. EOS Trans. AGU 84 (Fall Meet. Supl. Abstract), S41C-0107.
• 52)   Ando,  M.,  Tu,  Y.,  Kumagai,  H.,  Yamashita,  Y. and  Lin,  C.H. (2012) Very low-frequency earthquakes along the Ryukyu subduction zone. Geophys. Res. Lett. 39, L04303.
• 53)   Asano,  Y.,  Obara,  K. and  Ito,  Y. (2008) Spatiotemporal distribution of very-low frequency earthquakes in Tokachi-oki near the junction of the Kuril and Japan trenches revealed by using array signal processing. Earth Planets Space 60, 871–875.
• 54)   Matsuzawa,  T.,  Asano,  Y. and  Obara,  K. (2015) Very low frequency earthquakes off the Pacific coast of Tohoku, Japan. Geophys. Res. Lett. 42, 4318–4325.
• 55)   Ito,  Y. and  Obara,  K. (2006) Dynamic deformation of the accretionary prism excites very low frequency earthquakes. Geophys. Res. Lett. 33, L02311.
• 56)   Sugioka,  H.,  Okamoto,  T.,  Nakamura,  T.,  Ishihara,  Y.,  Ito,  A.,  Obana,  K. (2012) Tsunamigenic potential of the shallow subduction plate boundary inferred from slow seismic slip. Nat. Geosci. 5, 414–418.
• 57)   Nakano,  M.,  Hori,  T.,  Araki,  E.,  Kodaira,  S. and  Ide,  S. (2018) Shallow very-low-frequency earthquakes accompany slow slip events in the Nankai subduction zone. Nat. Commun. 9, 984.
• 58)   Kaneko,  L.,  Ide,  S. and  Nakano,  M. (2018) Slow earthquakes in the microseism frequency band (0.1–1.0 Hz) off Kii Peninsula, Japan. Geophys. Res. Lett. 45, 2618–2624.
• 59)   Baba,  S.,  Takeo,  A.,  Obara,  K.,  Matsuzawa,  T. and  Maeda,  T. (2020) Comprehensive detection of very low frequency earthquakes off the Hokkaido and Tohoku Pacific coasts, northeastern Japan. J. Geophys. Res. Solid Earth 125, e2019JB017988.
• 60)   Nakamura,  M. and  Kakazu,  K. (2017) Tidal sensitivity of shallow very low frequency earthquakes in the Ryukyu trench. J. Geophys. Res. Solid Earth 122, 1221–1238.
• 61)   Obana,  K. and  Kodaira,  S. (2009) Low-frequency tremors associated with reverse faults in a shallow accretionary prism. Earth Planet. Sci. Lett. 287, 168–174.
• 62)   Ito,  Y.,  Hino,  R.,  Suzuki,  S. and  Kaneda,  Y. (2015) Episodic tremor and slip near the Japan Trench prior to the 2011 Tohoku-Oki earthquake. Geophys. Res. Lett. 42, 1725–1731.
• 63)   Nishikawa,  T.,  Matsuzawa,  T.,  Ohta,  K.,  Uchida,  N.,  Nishimura,  T. and  Ide,  S. (2019) The slow earthquake spectrum in the Japan Trench illuminated by the S-net seafloor observatories. Science 365, 808–813.
• 64)   Tanaka,  S.,  Matsuzawa,  T. and  Asano,  Y. (2019) Shallow low frequency tremor in the Northern Japan Trench subduction zone. Geophys. Res. Lett. 46, 5217–5224.
• 65)   Ito,  Y.,  Hino,  R.,  Kido,  M.,  Fujimoto,  H.,  Osada,  Y. and  Inazu,  D. (2013) Episodic slow slip events in the Japan subduction zone before the 2011 Tohoku-Oki earthquake. Tectonophysics 600, 14–26.
• 66)   Kato,  A.,  Obara,  K.,  Igarashi,  T.,  Tsuruoka,  H.,  Nakagawa,  S. and  Hirata,  N. (2012) Propagation of slow slip leading up to the 2011 M_W 9.0 Tohoku-Oki earthquake. Science 335, 705–708.
• 67)   Araki,  E.,  Saffer,  D.M.,  Kopf,  A.J.,  Wallace,  L.M.,  Kimura,  T.,  Machida,  Y. (2017) Recurring and triggered slow-slip events near the trench at the Nankai Trough subduction megathrust. Science 356, 1157–1160.
• 68)   Yokota,  Y.,  Ishikawa,  T.,  Watanabe,  S.,  Tashiro,  T. and  Asada,  A. (2016) Seafloor geodetic constraints on interplate coupling of the Nankai Trough megathrust zone. Nature 534, 374–377.
• 69)   Yokota,  Y. and  Ishikawa,  T. (2020) Shallow slow slip events along the Nankai Trough detected by GNSS-A. Sci. Adv. 6, eaay5786.
• 70)   Heki,  K. and  Kataoka,  T. (2008) On the biannually repeating slow slip events at the Ryukyu trench, southwestern Japan. J. Geophys. Res. Solid Earth 113, B11402.
• 71)   Nishimura,  T. (2014) Short-term slow slip events along the Ryukyu trench, southwestern Japan, observed by continuous GNSS. Prog. Earth Planet. Sci. 1, 22.
• 72)   Brudzinski,  M.R. and  Allen,  R.M. (2007) Segmentation in episodic tremor and slip all along Cascadia. Geology 35, 907–910.
• 73)   Audet,  P. and  Burgmann,  R. (2014) Possible control of subduction zone slow-earthquake periodicity by silica enrichment. Nature 510, 389–392.
• 74)   Nakajima,  J. and  Hasegawa,  A. (2016) Tremor activity inhibited by well-drained conditions above a megathrust. Nat. Commun. 7, 13863.
• 75)   Boatwright,  J. and  Cocco,  M. (1996) Frictional constraints on crustal faulting. J. Geophys. Res. Solid Earth 101, 13895–13909.
• 76)   Obara,  K. and  Sekine,  S. (2009) Characteristic activity and migration of episodic tremor and slow-slip events in central Japan. Earth Planets Space 61, 853–862.
• 77)   Obara,  K.,  Matsuzawa,  T.,  Tanaka,  S.,  Kimura,  T. and  Maeda,  T. (2011) Migration properties of non-volcanic tremor in Shikoku. Geophys. Res. Lett. 38, L09311.
• 78)   Annoura,  S.,  Obara,  K. and  Maeda,  T. (2016) Total energy of deep low-frequency tremor in Nankai subduction zone, southwest Japan. Geophys. Res. Lett. 43, 2562–2567.
• 79)   Kano,  M.,  Kato,  A.,  Ando,  R. and  Obara,  K. (2018) Strength of tremor patches along deep transition zone of a megathrust. Sci. Rep. 8, 3655.
• 80)   Ando,  R.,  Takeda,  N. and  Yamashita,  T. (2012) Propagation dynamics of seismic and aseismic slip governed by fault heterogeneity and Newtonian rheology. J. Geophys. Res. Solid Earth 117, B11308.
• 81)   Uchida,  N.,  Takagi,  R.,  Asano,  Y. and  Obara,  K. (2020) Migration of shallow and deep slow earthquakes toward the locked segment of the Nankai megathrust. Earth Planet. Sci. Lett. 531, 115986.
• 82)   Obara,  K.,  Tanaka,  S.,  Maeda,  T. and  Matsuzawa,  T. (2010) Depth-dependent activity of non-volcanic tremor in southwest Japan. Geophys. Res. Lett. 37, GL43679.
• 83)   Wech,  A.G. and  Creager,  K.C. (2011) A continuum of stress, strength and slip in the Cascadia subduction zone. Nat. Geosci. 4, 624–628.
• 84)   Hirose,  H.,  Asano,  Y.,  Obara,  K.,  Kimura,  T.,  Matsuzawa,  T.,  Tanaka,  S. (2010) Slow earthquakes linked along dip in the Nankai subduction zone. Science 330, 1502.
• 85)   Husker,  A.L.,  Kostoglodov,  V.,  Cruz-Atienza,  V.M.,  Legrand,  D.,  Shapiro,  N.M.,  Payero,  J.S. (2012) Temporal variations of non-volcanic tremor (NVT) locations in the Mexican subduction zone: Finding the NVT sweet spot. Geochem. Geophys. Geosyst. 13, Q03011.
• 86)   Kato,  A.,  Iidaka,  T.,  Ikuta,  R.,  Yoshida,  Y.,  Katsumata,  K.,  Iwasaki,  T. (2010) Variations of fluid pressure within the subducting oceanic crust and slow earthquakes. Geophys. Res. Lett. 37, L14310.
• 87)   Holtkamp,  S. and  Brudzinski,  M.R. (2010) Determination of slow slip episodes and strain accumulation along the Cascadia margin. J. Geophys. Res. Solid Earth 115, B00A17.
• 88)   Nakamura,  M. and  Sunagawa,  N. (2015) Activation of very low frequency earthquakes by slow slip events in the Ryukyu Trench. Geophys. Res. Lett. 42, 1076–1082.
• 89)   Takemura,  S.,  Matsuzawa,  T.,  Noda,  A.,  Tonegawa,  T.,  Asano,  Y.,  Kimura,  T. (2019) Structural characteristics of the Nankai Trough shallow plate boundary inferred from shallow very low frequency earthquakes. Geophys. Res. Lett. 46, 4192–4201.
• 90)   Yabe,  S. and  Ide,  S. (2013) Repeating deep tremors on the plate interface beneath Kyushu, southwest Japan. Earth Planets Space 65, 17–23.
• 91)   Takemura,  S.,  Noda,  A.,  Kubota,  T.,  Asano,  Y.,  Matsuzawa,  T. and  Shiomi,  K. (2019) Migrations and clusters of shallow very low frequency earthquakes in the regions surrounding shear stress accumulation peaks along the Nankai Trough. Geophys. Res. Lett. 46, 11830–11840.
• 92)  Ministry of Education, Culture, Sports, Science and Technology (MEXT). (2013) Evaluation of occurrence potentials or subduction-zone earthquakes. https://www.jishin.go.jp/main/chousa/13may_nankai/nankai_gaiyou.pdf (in Japanese).
• 93)  National Research Institute for Earth Science and Disaster Resilience (NIED). (2016) Activity of deep low-frequency tremor in southwest Japan (November, 2015–April, 2016). Rep. CCEP (Report of The Coordinating Committee for Earthquake Prediction, Japan) 96, 363–369.
• 94)   Roeloffs,  E.A. (2006) Evidence for aseismic deformation rate changes prior to earthquakes. Ann. Rev. Earth Planet. Sci. 34, 591–627.
• 95)   Nadeau,  R.M. and  Guilheim,  A. (2009) Nonvolcanic tremor evolution and the San Simeon and Parkfield, California, Earthquakes. Science 325, 191–193.
• 96)   Shelly,  D.R. (2010) Migrating tremors illuminate complex deformation beneath the seismogenic San Andreas fault. Nature 463, 648–652.
• 97)   Kato,  A. and  Nakagawa,  S. (2014) Multiple slow-slip events during a foreshock sequence of the 2014 Iquique, Chile M_W 8.1 earthquake. Geophys. Res. Lett. 41, 5420–5427.
• 98)   Kano,  M.,  Kato,  A. and  Obara,  K. (2019) Episodic tremor and slip silently invades strongly locked megathrust in the Nankai Trough. Sci. Rep. 9, 9270.

Profile

Kazushige Obara was born in 1959 in Sendai, Miyagi Prefecture. After he received his master’s degree from Tohoku University in 1985, he served the National Research Institute for Earth Science and Disaster Prevention (NIED) as a government researcher of the Prime Minister’s Office of Japan. His work was related to the maintenance and data analysis of the deep borehole seismic observatories in the Tokyo metropolitan area and Kanto–Tokai crustal activity observation network. He received his PhD degree in 1992 from Tohoku University for his study on scattering of seismic wave propagation and worked as a visiting research fellow at MIT for one year in 1993. After the Kobe earthquake in 1995, he participated in the project for constructing a nationwide high-sensitivity seismograph network (Hi-net). He was appointed as the head of the Hi-net management laboratory in 2001 and the director of the seismic observation data center in 2006. When working on Hi-net data, he discovered some types of slow earthquakes: low-frequency tremor, very-low-frequency earthquakes, and short-term slow slip events. In 2010, he moved to the Earthquake Research Institute (ERI), the University of Tokyo, as a professor and was elected as the Director of ERI from 2015 to 2019. He is the leader of the project “Science of Slow Earthquakes” funded for five years from 2016 via the Japan Society for the Promotion of Science, Grant-in-Aid Scientific Research on Innovative Areas. For his accomplishment, he received the Ministry of Education, Culture, Sports, Science and Technology (MEXT) Commendation in 2003, Inoue Award in 2008, American Geophysical Union (AGU) Fellow in 2013, AGU Gutenberg Lecturer in 2015, and Seismological Society of Japan (SSJ) Prize in 2018. He is currently the President of SSJ.